Where Men Don't Work

Readers are going to have to endure quite a lot of map-based posts while I figure out the mechanics of turning census data into meaningful information.  In this post, I explore the geography of the male employment/population (E/P) ratio via county-level data (which I've just about mastered how to visualize).  County level data is enough for us to roughly distinguish urban areas from rural areas, but not to see individual neighborhood demographics (that requires census tract level data, and I haven't managed the scaling challenges of dealing with that in Google Earth just yet).

All the data in this post are based on looking at the ratio of men over the age of 16 who are in the labor force but not unemployed.  This is what is most readily available from the census, but notice that it confounds college attendance and retirement with underemployment - issues we will have to try to deal with in the future.  The color scale runs from 25% of men employed (white) to 75% employed (fully saturated blue). The data are from the 2000 US census (from the American Fact Finder website, census summary file 3).  So again, this is dealing with the country as it was at the height of the tech-boom, in a cylical high of employment.

The first map just shows a flat map of the whole country on this scale.  Like all images in this post, this can be clicked to get a pretty big (O(1MB)) version in a separate window for easier study.

As you can see, there is huge geographical variation in the fraction of men working: we pretty much touch both the lower and the upper boundaries of the 25%-75% range.  Areas of particularly low employment-population for men include the south-west (especially Nevada/Arizona/New Mexico), the interior South, and Appalachia.

It's hard in the above post to get a really clear idea of where the weight of the population is - are we dealing with only 25% of a large population working, or just of a small population.  To get a handle on this I created some visualizations in which I use the height variable to denote county population density.  More precisely, I created a synthetic landscape in which the translucent blue floats at 5000m plus ten times the population density of the county (in persons/sq mile) above sea level.  Thus the apparent volume of each county block (height times area) corresponds to the population of that county.  Meanwhile, the blue scale continues to represent male employment-population ratio (0.25-0.75).  (The 5000m is just to guarantee that the artificial topography is always at least slightly above the real topography.)

Here's how it looks from somewhere over Baja California looking north up the Pacific coast:

You can see the urban areas - sprawling LA/San Diego, dense San Francisco with the less dense environs of the greater Bay Area, and then the Portland and Seattle metro areas off in the distance up the coast.  Clearly, the urban areas mostly have higher male E/P ratios (though Los Angeles county looks like a pretty big slab of mediocre E/P ratio). Rural regions vary wildly, with some having high E/P (see northern Utah and southern Idaho), and others having much less.  However, for the most part, the very lowest E/P ratios seem to come in low density areas.

Next, we have another view of the same visualization from somewhere over New Mexico looking out over Texas and the Gulf Coast.

In this region, the more dense counties for the most part have a higher fraction of their menfolk employed.  Finally, we have a view from somewhere over the southern US looking north-east towards New York (the highest density county in the country), but taking in a sweep from the population centers of the upper midwest, round through the eastern seaboard and then down as far as Atlanta.

I'm quite proud of that picture.  The mountain ranges of civilization are pretty striking.

Finally, here is more traditional scattergram of the data, showing male E/P ratio versus population density (logarithmic scale) for the counties mapped above. I have added a moving average of sixty counties (ordered by population density) to show a smoothed version.

Population density is not strongly explanatory of the ratio of men that are working, but the data do show some structure, with outer suburban densities of about 500-1000/square mile having the highest E/P ratios, while urban areas have slightly lower E/P.  Meanwhile the poorest E/P ratios occur at rural densities of around 20-50/sq.mi.  In particular, almost all the values of E/P below 1/3 occur in the range of 7 -100/sq. mi.  But then in very sparsely populated areas (below 7/sq. mi.), a larger fraction of the few men there are working.

In future posts, I will try to take up the correlation of male E/P ratio with other variables of interest.